The birth order puzzle.

Abstract

Studies relating intellectual performance to birth order report conflicting results, some finding intellectual scores to increase, others to decrease with birth order. In contrast, the relationship between intellectual performance and family size is stable and consistently replicable. Why do these two highly related variables generate such divergent results? This birth order puzzle is resolved by means of the confluence model that quantifies the influences upon intellectual growth arising within the family context. At the time of a new birth, two opposing influences act upon intellectual growth of the elder sibling: (a) his or her intellectual environment is "diluted" and (b) he or she loses the "last-born's handicap" and begins serving as an intellectual resource to the younger sibling. Since these opposite effects are not equal in magnitude, the differences in intellectual performance among birth ranks are shown to be age dependent. While elder children may surpass their younger siblings in intellectual performance at some ages, they may be overtaken by them at others. Thus when age is taken into consideration, the birth order literature loses its chaotic character and an orderly pattern of results emerges.

The confluence model: Parameter estimation for six divergent data sets on family factors and intelligence

“…set (Breland, 1974), the birth order effect is clearly negative, following the typical pattern of findings (Zajonc et al, 1979). However, in another data set (Davis, Cahan, & Bashi, 1976), the birth order effect is negative only for small families.…”

“…Another component of these variations, however, depends on the age of testing. It was shown in a recent publication (Zajonc et al, 1979), that the age of testing has a significant influence over birth order effects. When a new child enters the family, two significant changes are assumed to occur in the process of intellectual development of the older child.…”

“…The initial M for a child at birth was assigned the value zero. And as previously (e.g., Markus & Zajonc, 1977;Zajonc, et al, 1979) the parental M's for each family size were set to a constant value of 19 for the Dutch, French, Scottish, and American data. Since information about parental education was available for the two Israeli samples, estimates of the actual parental M's were made accordingly.…”

“…The two components, at and ~,t, are therefore expressed as weighted yearly increments of this sigmoid function, Af(t) : (1 -e -k¢~) -(1 -e -k~l-ll2) =e -*2(~-lF -e-k~:~: The above terms wl and w2 are weights associated with the two components, f(r) = 1 -e -t 2~'2where r is the age of the adjacent younger sibling, and Lt is the last child index which is equal to zero for children who at age t are not followed by a younger sibling and equal to one otherwise. A more detailed and extensive psychological interpretation of the specific features of the confluence model is given elsewhere (Zajonc et al, 1979). Suffice it to point out here that the weights Wl and w2 represent metric contributions of the component sources of influence upon intellectual growth, and need to be estimated from the data.…”

“…A reparametrization of this model was subsequently employed to explicate the conflicting results reported in the birth order literature (Zajonc, Markus, & Markus, 1979). This paper reports more stringent and extensive tests of the confluence model than have been carried out thus far.…”